Everything is a Freaking DNS problem

As I was already up since yesterday 0500 , it was dinner with Sven , Robin and some other conference visitors at a Turkish Buffet place , after which we headed to what seemed to be a great bar where they failed to serve us while waiting for over 10 minutes, so we moved on to another place. and then to be "early"

After walking around a bit in Copenhagen and looking for a bus stop to go to the university I managed to bump into Wim & Co who offered me a ride to the IT University. Where I was almost in time for the first talk by
Jan Wieck about Slony-I, A master to multiple slaves-replication system for PostgreSQL
Given my recent MySQL MultiMaster setups I was fairly interested where PostgreSQL is at today.

Jan started out with explaining where he used replication the most,
For backups and Specialized services so he could offload long running and intrusive reporting tools to an isiolated server.

While going over the history of Slony he also mentioned eRserver, first written in Perl later rewritten in Java and that was a ... Well lets just say that memory usage wasn't really ideal.

The presentation covered different potential replication scenarios and the problems one could run into.
No sign however of multimaster replication. Jan Wieck even told us he had no plans to implmenent multimaster replication
at all at the moment. To me MultiMaster means that I can move my database connection together with my application service in a HA setup. I don't need to wory about the possibility of writing in a slave and breaking replication as the slave is also a master and the other node Will catch up.

Next up was a talk about openID, which made me realize a couple of things about it.. all off that in a separate post :)

The University restaurant was open and you could pay for your Chilli Con Carne to a really Grumpy cashier which brings me right to the next talk I followed.. A Developers Guide to Grumpy Old Sysadmins however, the majority of people in the room were Sysadmins , or people with a mixed role doing both development and sysadmin work.

You don't let your developers even close to your production systems. He then went on to read a fairly big part of "Over Clocked, Stories of the Future Present " by Cory Doctorow , which I've promptly put on my holliday reading list :

So what's it like being a sysadmin ? You get calls in the middle of the night because a system breaks,
When you work you .. people complain when things go wrong.. they don't cheer when things go right or when you have done your job correctly

Reminds me of this situation at a customer where the Cisco people that fail to automate their work and get to travel around the qworld to do their work using their serial console , where as the Linux platform team has automated their work so far they get to stay home and watch the machines boot then do a quick check over ssh to see if they actually work.

Anyhow.. the talk really wasn't focusing on how developers could cooperate better with sysadmins, apart from a couple of general tips so it really missed its goal.

I popped into the Lightning talk about Ubuntu on EEE talk , hoping to learn something, apart from 1 url that I should read I guess it was the otherway around , it's usually a bad sign when a speaker starts every 2 slides with "I haven't tested this myselve but " ...

I've just placed the presentation I gave both yesterday at the Open Source Days in Copenhagen , and last week in Zurich at the Open Expo , about Open Source Virtualization online.
The presentation is based on a series of articles I wrote earlier this year for Virtualization.com

You can download it here

The presentation covers a fairly complete overview of what's around in Open Source Virtualization tools and and their Management frameworks.

I will be giving the same presentation again at the end of the month at T-Dose in Eindhoven, The Netherlands

I'm expecting different people not using Dopplr to come back to me next week and tell me they were in Denmark also , as Copenhagen is the place to be this week(end)

I was late for the first talk on how to automate a large scale school environment with gentoo. So I missed the rationale for the development of DIPO, and I didn't want to bother the speaker with questions he had already explained in his earlier slides, however .. it spelled "reinventing the wheel" all over again.

The second speaker apologised for being late on schedule as the first speaker had to compile everything. Nigel Kersten from from Google talked about Puppet, need I say more.. it was great :)

So lunch is really really early here as from 11u30 on , however I didn't mind as I was up since 5 , so I actually was pretty hungry :)

After lunch 3 talks and a Keynote were scheduled. In order not to be late for my own talk I headed into the Open Solaris Storage talk ... , not really interesting :(

Funny quote from the talk however "The clients are still Linux, but the server is already solaris"...funny thing to say ... certainly for a dying breed.

My own talk was fairly well attended, people were sitting on the stairs in the Auditorium and there was a lot of Q&A after.

As said in my talk last week in Zurich I don't think
Sun has a really clear message with what they are doing with xVM etc
So the talk aftermine should have cleared that out shouldn't it.
Well it didn't , all I got back from it was .. "We ported Xen to Solaris" , and "Our demo doesn't work because of a funky dns problem."

So Robin Rowe is keynoting about about Linux at the Movies.. and tonight it's off to the Social event .. well one of them :)

Seth is having the same problems I have, sharing your calendar with different groups of people. Where some of these groups have their own system in place already.

Only he has a partial solution for it.

I'm looking for a step further.. it could probably be called Calendar Delegation.
I kind of would like to have 1 calendar that I manage and based on from where you see that calendar you get detailed information or just free busy. So my favorite customer would see if I`m at their office or not
and if I`m having meetings at their office, however whem I`m not at their office they would just see a busy.

My collegues should however be able to see that I`m at a customer but not having any meetings there.

And I should only have to manage everything in one place while for my customers they should just be able to see it in their calendar solution flavor of the week.

A man can dream right ? :)

As mentioned earlier , Packt sent me their Zenos Core Network and Systems Monitoring by Michael Badger to review.

So I spend the past couple of weeks trying to crawl my way trough it. Not that there was any problem with the book, more with my schedule :) I was hoping to finish it some time ago but time was working against me, even last week when my flight back from Zurich was a bit shorter than planned I couldn't finish reading the book.

But yesterday I managed, Michael managed to write a good book on a hot topic during what was probably one of the most busy of his life.

In 8 chapters Michael manages to explain the different aspects of Zenoss fairly well, he even touches the creation of ZenPacks and extending Zenoss in general also.

It was a fairly good read and showed me a couple of new insights into Zenoss.

However although Zenoss is heavily SNMP based there is little information on SNMP in general and MIB more in detail ? People really new to systems monitoring might want to have a couple of more pages on these topics.

For those not familiar to Zope the mentioning of TALES expressions might also raise some eyebrows, luckily there has been a full Appendix dedicated to it.

Time permitting I would have hoped he could of gone a bit more in detail on creating a ZenPack that did a bit more than checking the status of a page. Looking into the status of an Apache or MySQL or other relevant daemon , or the Squid internal statistics

But if you are new to Zenoss and you want to have a quick headstart into monitoring your infrastructure with it this book is a really really good start.

There's a free preview Chapter at Scribd.com and also at Packt

I misplaced my notes, but I must say that day 2 of the openExpo was pretty interresting.. some talks that when first looking at them weren't that interresting turned out to be extremely timely.

I however got a couple of nice pictures from the container that Harald already mentioned

Tarry pointed us to an article questioning if the internet is ready for cloud computing.

The short summary,
the biggest challenge is DNS.

Is DNS going to be secure enough, scalable enough etc.

Don't say I didn't warn you .. "Everything is a Freaking DNS Problem !"

Yesterday apart from Free Beer, the social event also featured a couple of 5 minute lightning talks.
Tobias Oetiker took the first slot, his 5 minute talk was incredible. In 5 minutes he went trough 80 something slides , really remined me of the St Peter talk about Jabber and Security a couple of years ago at Fosdem. I after 2 talks I realised that I could strip down my OLS openQRM talk and give it in 5 minutes too .. so I did :) Awesome.. 5 minutes is really really really short :)
But I managed .. I even got questions at the end :)

Definitely a good concept to let people decide on giving a talk at such a short notice :)

Everybody is in Zwitserland these days .. even the Userfriendly crowd , so am I :)

I'm in Winterthur for the OpenExpo Zurich, a less eventfull flight than my last conference trip brought me to my hotell where I crashed and after a short walk to the conference site I managed to catcht the last part of Bruce Perens talk

He talked about Spies using Open Source, he talked about Casinos in
Vegas wanting interoperability and asking him for help.

Bummer the Mozilla talk was in German, so I left ..

Altough asked to give the Open Source in the Telco industry talk in English it was in German too .I tried to follow ..failed .. to bad as most of the work I do is in the Telco industry .

Enrico Zini talked about Debian Diveristy and the clear conclusion is that
Debian people are weirdos :) Some of them insist in wearing Kilts a bit more thatn the average male on this planet. Some of them meet their wife at Debian Conferencesm others go to debconf on honeymoon.. they go on holiday to space..
start contributing less, build their own meta distro and take away a lot of the less experienced people, which was a good thing.

After Enrico Max Spevack took the floor to talk about Fedora, he was the second in line to apologise for not doing his talk in German. Weird..
I'm in doubt in which language I'll give mine . Dutch of French .
There are significantly less people in this talk compared to the Debian talk

No real info on the recent security hickup however, I got a Fedora T-Shirt for asking though :)

So earlier this week I ran into the weirdest problem with Linux-HA. Heartbeat was happily adding an IP address as an active resource so one of my nodes when needed, but upon removal it failed to remove the IP from the stack. Further debugging learned that the Heartbeat scripts claimed the ip wasn't on the actual stack.

It was.. but it the output from ifconfig was different from what it expected it to be.

Heartbeat checks the output of ifconfig and expects to find the IP address it added itselve to be on a :0 or similar interface. Now ifconfig only seems output 8 characters for the interface name Which means that when you have an interface called eth0:0 the output perfectly lists it and heartbea
t is smart enough to remove the ip again when the node goes to standby. If however you have a vlan with 3 digits on a bond interface Heartbeat will add :0 to bond0.129 , the Heartbeat resource will add the ip address perfectly but opon checking all the :0 interfaces the bond0.129:0 interface won't be parsed as ifconfig outputs it as bond0.129 , hence resulting in a potential painfull situation where 2 nodes still share an IP address.

So where's the actual problem ifconfig, or heartbeat, I'd say both, but the easiest fix will be in Heartbeat, afterall there are other preferred ways of adding an ip addres to an interface. ip addr add comes to mind :)

So I filed a bug report :)